一项最新研究声称，AI加密挖矿网络Pearl在未进行任何可验证AI计算的情况下，消耗了大量的电力和GPU资源。该网络据称使用了相当于32万块RTX 3090级GPU的算力，消耗112兆瓦电力，来进行随机矩阵乘法运算，仅仅模仿AI数学。据报道，此举已导致vast.ai等平台上的GPU租金上涨38%，影响了独立研究人员。 AI
影响 “有用工作证明”方案的广泛采用可能会推高GPU成本，并从真正的AI开发中转移计算资源。
排序理由 一项研究声称“有用工作证明”系统并未执行有用工作。
在 Mastodon — fosstodon.org 阅读 →
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
A preprint claims Pearl’s AI mining network consumes 320,000 GPU-equivalents and 112 MW while producing no verified useful AI computation.
AI cryptomining network's 320,000 RTX 3090-class GPUs allegedly burn 112 megawatts of power on ‘zero useful AI computation’ — GPU rental costs jump 38%, but Pearl’s cards are doing random… A preprint claims Pearl’s AI mining network consumes 320,000 GPU-equivalents and 112 MW whi…
♻️ AI cryptomining network's 320,000 RTX 3090-class GPUs allegedly burn 112 megawatts of power on ‘zero useful AI computation’ — GPU rental costs jump 38%, but Pearl’s cards are doing random matrix math, study claims https://www. tomshardware.com/tech-industry /artificial-intelli…
AI cryptomining network's 320,000 RTX 3090-class GPUs allegedly burn 112 megawatts of power on ‘zero useful AI computation’ — GPU rental costs jump 38%, but Pearl’s cards are doing random matrix math, study claims https://www. tomshardware.com/tech-industry /artificial-intelligen…